Suppression of Metastases Using a New Lymphocyte Checkpoint Target for Cancer Immunotherapy

The induction and progression of a protective adaptive immune response is a tightly controlled process, mediated by multiple immune checkpoints, and involves engagement of activating, costimulatory signals and the avoidance of negative or coinhibitory signals (1). A group of receptors known as the immunoglobulin superfamily play a central role in controlling lymphocyte-driven immune responses, with the CD28/cytotoxic T-lymphocyte antigen 4 (CTLA-4):B7.1/B7.2 and programmed cell death protein 1 (PD-1):programmed death ligand 1 (PD-L1)/PD-L2 receptor:ligand interactions the most extensively studied (2, 3). One function of these checkpoints is to protect the host against harmful immune responses, be it preventing immune responses against self or dampening overactive responses against foreign pathogens (4, 5). Although this is necessary in the prevention of autoimmunity, immune checkpoints are a barrier to successful immunotherapies targeting malignant cells, where the majority of surface molecules expressed are self-antigens. Malignant cells can also actively suppress immune responses directed against them, in part through the upregulation of inhibitory ligands, such as PD-L1, at the tumor site (6). Therapies aimed at overcoming immune tolerance by blocking the interaction of inhibitory ligands with immune checkpoints have the potential to enhance T-cell antitumor immune responses (1). Indeed, blocking antibodies against CTLA-4, PD-1, and PD-L1 have demonstrated unprecedented efficacy for an immunotherapy against a number of advanced human cancers (7–11).

Natural killer (NK) cells are part of the innate lymphocyte family and play a prominent role in controlling early tumor growth and the spread of metastases through cytotoxic activity and the release of inflammatory cytokines (12). As with T cells, the function of NK cells is strongly regulated, partly through the binding of activating and inhibitory receptors expressed on these cells (13). Activating receptors such as natural cytotoxicity receptors (NCR), natural-killer group 2, member D (NKG2D), or CD226 (DNAM-1) can recognize and interact with a range of ligands upregulated by pathogens or cellular stress, stimulating NK-cell cytotoxicity and secretion of proinflammatory cytokines such as IFNγ (14). Opposing this activity, inhibitory receptors expressed on NK cells interact with ligands that protect normal cells from NK-cell cytotoxicity (15). This family of receptors includes killer cell immunoglobulin-like receptors (KIR) and leukocyte immunoglobulin-like receptors (LIR) that interact with MHC class I and MHC class I related molecules.

More recently, a group of immunoglobulin superfamily receptors that interact with ligands of the nectin and nectin-like (NECL) family have been described as having a significant role in altering NK- and T-cell functions (16). These include CD226 (17), CD96 (TACTILE; ref. 18), T-cell immunoglobulin and ITIM domain (TIGIT; refs. 19, 20), and class-I restricted T cell-associated molecule (CRTAM; ref. 21). CD226 and TIGIT interact with a pair of common ligands, CD155 (NECL5; PVR) and CD112 (NECTIN2; PVRL2), that are often highly upregulated on tumor cells (19, 22). TIGIT has also been shown to interact with the ligand CD113 (PVRL3; ref. 19). CD226 and TIGIT appear to have counterbalancing activity on NK cells, with CD226 acting as an activating receptor and TIGIT as an inhibitory receptor (23). In vitro, CD226 is vital for NK-cell cytotoxicity against tumor cells (24, 25) and also plays a critical role for in vivo tumor immunosurveillance (24, 26). TIGIT, however, has an ITIM motif and has been predicted to play a role in limiting NK-mediated tissue damage in a similar manner to Ly49 or KIR binding to MHC class I (27). Confirming this, the binding of TIGIT to CD155 was shown to reduce in vitro NK-cell IFNγ production and cytotoxicity (28, 29).

Despite being cloned 20 years ago (18), little is known about CD96, the other Ig family member, other than that it also interacts with the CD155 ligand (30, 31). CD96 expression is largely constrained to NK cells and CD8⁺ and CD4⁺ T cells in humans (18). In mice, CD96 is found on NK cells, CD8⁺ T cells, CD4⁺ T cells, γδ T cells, and natural killer T (NKT) cells (32). Similar to CD226 and TIGIT, the major ligand of CD96 is CD155, but it has also been shown to associate with CD111 (NECTIN1) to promote NK- and T-cell adhesion (31, 33). In a recent study, we demonstrated that CD96 competes with CD226 for CD155 binding and limits NK-cell functions by direct inhibition (32). As a result, Cd96^−/− mice displayed hyperinflammatory responses to the bacterial product lipopolysacharide (LPS) and showed greater protection against B16F10 lung metastases and the development of carcinogen-induced cancers. These data provided the first description of the ability of CD96 to negatively control cytokine responses by NK cells. Thus, blocking CD96 may have applications in pathologies where NK cells play an important role, particularly in suppressing tumor metastases. In this study, we have evaluated anti-mouse CD96 mAbs as a novel cancer immunotherapy, alone and in the context of conventional therapies, such as surgery, chemotherapy, and immune checkpoint blockade, particularly with anti–PD-1 or anti–CTLA-4, against experimental or spontaneous metastases. Our data demonstrate the general utility of anti-CD96 mAb in enhancing NK-cell IFNγ-dependent effector function, independently of antibody-dependent cell-mediated cytotoxicity (ADCC), against experimental and spontaneous metastases.

Given the known constitutive expression of CD96 on mouse NK cells, we determined whether CD96 might affect NK cell–dependent antitumor immunity using different models of experimental lung metastases, where NK cell–mediated control has been previously demonstrated (34–36). B16F10, a melanoma cell line, expresses high levels of CD155 and CD112, and, consistent with previous reports, Cd226^−/− mice had increased lung metastases compared with wild-type (WT) mice (26), whereas Cd96^−/− mice had significantly fewer lung metastases (Fig. 1A; ref. 32). A similar pattern of results was obtained with 3LL lung carcinoma (Supplementary Fig. S1A), RM-1 prostate carcinoma (Supplementary Fig. S1B), and LWT1 melanoma cells (Supplementary Fig. S1C). We have previously demonstrated the ability of an anti-mouse CD96 mAb (clone 3.3) to enhance NK-cell IFNγ production (32). Given the phenotype of the Cd96^−/− mice with respect to metastases control, we next wished to establish whether a blocking antibody to CD96 could also limit experimental lung metastases. Initially, using the B16F10 metastasis model, we conducted a dose titration experiment with anti-CD96 mAb (clone 3.3; ref. 32; Fig. 1B). Doses as low as 20 μg of anti-CD96 mAb given on days 0 and 3 significantly suppressed B16F10 lung metastases compared with 250 μg control Ig (cIg), with optimal anti-CD96 mAb effects obtained at a dose of 250 μg. Having optimized the dose, we then showed that 250 μg of anti-CD96 mAb given on days 0 and 3 resulted in a similar reduction in mice bearing 3LL (Fig. 1C) or RM-1 metastases (Fig. 1D). We also tested the ability of another anti-CD96 mAb (clone 6A6) that blocks CD96–CD155 binding (31) to reduce B16F10 lung metastases. Both clones of anti-CD96 mAbs reduced lung metastases compared with cIg, but clone 6A6 had more potent antimetastatic activity than clone 3.3 (Fig. 1E). Both anti-CD96 mAbs bound to recombinant mouse CD96 protein with high affinity (clone 3.3, Kd = 186 pmol/L; 6A6, Kd = 8.3 nmol/L). Overall, these data demonstrated that the protection against lung metastases observed in Cd96^−/− mice was also achieved with two different blocking anti-CD96 mAbs.

Using the B16F10 melanoma lung metastasis model, we next examined the mechanism by which anti-CD96 mAb mediated its antimetastatic activity. Suppression of metastases by anti-CD96 was clearly lost in the absence of NK cells or following neutralization of IFNγ, but not significantly diminished in the absence of host T cells (anti-CD4/anti-CD8β; Fig. 2A). These results were further supported by experiments in perforin-deficient mice (Pfp^−/−) where anti-CD96 mAb retained activity against B16F10 metastases, but lost activity when IFNγ was neutralized (Fig. 2B). Further analysis revealed that anti-CD96 mAb was also effective at reducing metastases in mice heterozygous for CD96, but only partially effective in mice lacking IL12p35 (Supplementary Fig. S2A). Antimetastatic activity was also maintained in Rag1^−/− or Rag2^−/− mice lacking T and NKT cells, whereas activity was lost in Rag2^−/−;Il2rg^−/− or NKp46^Cre/WT;Mcl1^fl/fl mice lacking NK cells (Supplementary Fig. S2B–S2D). Importantly, in the B16F10 lung metastasis model, anti-CD96 suppression of metastases was also shown to be dependent upon CD226 (Fig. 2C). Although differences in the number of metastases between WT mice and mice lacking all activating Fc receptors (Fceγ) or mice specifically lacking FcγRIII or FcγRIV were observed (Fig. 2D), the effect of anti-CD96 mAb activity did not appear to be Fc receptor–dependent, as suppression of metastases was seen with antibody treatment in all Fc receptor knockout mice. These data are consistent with an antagonist role of the rat IgG1 anti-mouse CD96 mAb in enhancing NK-cell IFNγ production and antimetastatic activity in the absence of any critical ADCC or antibody-dependent cellular phagocytosis.

TIGIT has recently been implicated as a marker of T-cell exhaustion in preclinical mouse models and patient cancers (37, 38). TIGIT has also been shown to limit NK cell–mediated cytotoxicity and IFNγ production (28, 29), and so we investigated if Tigit^−/− mice were resistant to experimental lung metastases and if CD96 blockade had an additional effect in Tigit^−/− mice. WT or Tigit^−/− mice were injected i.v. with B16F10 cells while also receiving cIg or anti-CD96 mAb on days 0 and 3. Tigit^−/− mice had no significant difference in lung metastases compared with WT (Fig. 3A), whereas blocking with anti-CD96 mAb reduced lung metastases in WT and Tigit^−/− mice. Interestingly, blocking CD96 in Tigit^−/− mice was able to further reduce the number of B16F10 metastases compared with treatment in WT mice. A similar result was also obtained in RM-1–bearing Tigit^−/− mice treated with anti-CD96 mAb (Fig. 3B). We also studied the role of TIGIT and CD96 in controlling spontaneous lung metastases in an adjuvant setting. Postoperative treatment approaches are very relevant because many patients have high risk of recurrence after primary tumor resection. The EO771 mammary carcinoma cell line was inoculated orthotopically into the mammary fat pad of WT or Tigit^−/− mice. Tumors were allowed to establish for 16 days before being resected, with treatment commencing on days 17, 21, 25, and 29 with cIg or anti-CD96 mAb. On day 35 following tumor inoculation, lungs were removed from mice and metastatic colonies counted. Again, no difference in metastases was observed between WT and Tigit^−/− mice; however, treatment with anti-CD96 mAb further reduced the level of metastases in the Tigit^−/− mice compared with WT mice (Fig. 3C). Cd96^−/− mice also demonstrated reduced metastases compared with WT mice with the result unchanged by treatment with anti-CD96 mAb (Supplementary Fig. S3). These results demonstrate targeting CD96 and TIGIT to be an effective combination to enhance protection against experimental and spontaneous metastases.

Although anti-CD96 mAb potently suppressed experimental and spontaneous metastases as a monotherapy, recent clinical trials combining immunotherapies have shown significant improvements in clinical outcomes compared with monotherapy treatments (9). Anti–CTLA-4 is a prototypic checkpoint blockade mAb, and its human equivalent, ipilimumab, has been FDA-approved for the treatment of advanced malignant melanoma (7). Anti–PD-1 (targeting T cells; ref. 10) and anti–PD-L1 (targeting tumor-expressed ligand; ref. 8) appear more effective and safer than targeting CTLA-4. Recent results for both nivolumab (10) and pembrolizumab (11) suggest that anti–PD-1 may be employed in a range of different cancer types in the future, with notable activity in melanoma. However, the therapeutic effect of these blocking approaches against metastases at distant sites is less addressed. Given this current status of cancer immunotherapy in melanoma, we conducted a series of experiments first using the B16F10 experimental lung metastases model, to compare the activity of anti-CD96 mAb alone and in combination with these contemporary agents.

In mice bearing B16F10 lung metastases, anti-CD96 mAb was more effective than either anti–CTLA-4 or anti–PD-1 as a monotherapy, with anti–CTLA-4 and anti–PD-1 displaying very modest antimetastatic activity compared with cIg (Fig. 4A). Interestingly, combinations of anti–CTLA-4/anti-CD96 (P= 0.0147) and more prominently anti–PD-1/anti-CD96 (P < 0.0001) were more effective than anti-CD96 alone (Fig. 4A). A very similar pattern of single-agent and combination therapy activity was also observed in mice bearing RM-1 lung metastases (Fig. 4B). Statistical evaluation of the interactions between anti-CD96 and anti–CTLA-4 or anti–PD-1 indicated an additive effect at reducing metastases of both antibody combinations (Supplementary Fig. S4A and S4B). We also evaluated if treatment with these mAbs alone or in combination could enhance the survival of mice bearing B16F10 or RM-1 lung metastases (Supplementary Fig. S5A and S5B). In concordance with the observed reduction in metastases, anti-CD96 was the most potent monotherapy at prolonging survival of mice, whereas anti-CD96 combined with anti–CTLA-4 or anti–PD-1 enhanced survival compared with anti-CD96 alone. We next examined whether the best combination (anti-CD96/anti–PD-1) remained effective in the B16F10 experimental lung metastasis model when treatment was delayed (days 5, 7, and 9; Fig. 4C). Clearly, both anti-CD96 alone and in combination with anti–PD-1 were less effective than when given early, but significant suppression of metastases by the combination, beyond each single therapy, was still noted.

The 4T1.2 mammary carcinoma is a spontaneous, highly metastatic tumor and, by employing surgical resection of the primary mammary gland tumor, represents a well-characterized and excellent model for mimicking human metastatic disease (39). Using this model, we examined in the context of surgery the adjuvant use of anti-CD96 mAbs alone or in combination with anti–CTLA-4 or anti–PD-1 mAbs (Fig. 4D). The median survival of mice treated with cIg immediately after surgery (day 20) was 35 days from tumor inoculation. Survival was somewhat enhanced by anti-CD96 alone (median survival, 54 days), anti–CTLA-4 (median survival, 49 days), and anti–PD-1 (median survival, 49 days). The combination of anti-CD96 with anti–CTLA-4 or anti–PD-1 enhanced survival compared with each agent as a monotherapy, extending median survival to 70 days and 74 days, respectively. The chemotherapy doxorubicin (DOX) has been shown to enhance antitumor activity by engaging the immune system (40). We therefore also examined the impact of DOX in combination with immune-enhancing antibodies. Although DOX combined with anti–CTLA-4 slightly improved survival compared with anti–CTLA-4 alone (median survival, 53 days vs. 49 days for CTLA-4 alone), the combinations of DOX with anti–PD-1 or anti-CD96 gave the greatest improvement in survival compared with the use of each agent as a monotherapy (median survival: CD96, alone 49 days vs. 73 days in combination; anti–PD-1, alone 49 days vs. 76 days in combination; Supplementary Fig. S6A). We also observed a significant decrease in the number of spontaneous 4T1.2 metastases in mice treated with anti-CD96, anti–CTLA-4, or anti–PD-1, with an additional reduction achieved when anti-CD96 was used in combination with anti–CTLA-4 or anti–PD-1 (Supplementary Fig. S6B). The 4T1.2 cells were shown to express high levels of CD155 and negligible levels of CD112 (Supplementary Fig. S6C). Overall, these data indicate the utility of anti-CD96 mAb alone or in combination with anti–CTLA-4, anti–PD-1, or DOX as an adjuvant therapy to control both experimental and spontaneous metastases.

The combination of anti-CD96 and anti–PD-1 mAbs gave the best protection against experimental and spontaneous metastases, so we investigated the mechanism of action of this combination in detail. Mice were injected i.v. with B16F10 cells while also receiving cIg, anti–PD-1 alone, anti-CD96 alone, or in combination on days 0 and 3. On day 6, mice were euthanized, and single-cell suspensions of lungs were generated and analyzed by flow cytometry. Treatment with all therapies, but notably anti-CD96 mAb alone or in combination with anti–PD-1, led to a significant increase in the number of infiltrating CD45.2⁺ immune cells, including NK cells within the lungs of these mice (Fig. 5A and B). NK-cell frequency was unaltered by anti-CD96 mAb therapy as an overall increase in both CD4⁺ and CD8⁺ T-cell numbers was also observed in these treatment groups (data not shown). Given this observation, we next determined if T cells were important in reducing metastases with the anti-CD96/anti–PD-1 combination. Mice bearing B16F10 lung metastases treated with anti-CD96/anti–PD-1 were depleted of either CD4⁺/CD8⁺ T cells or NK cells, and we showed that NK cells, but not T cells, were critical for the antimetastatic effect of this combination (Supplementary Fig. S7).

To assess if NK-cell function was also improved in these mice, lung single-cell suspensions were cultured for 24 hours, and IFNγ production within individual NK cells and levels secreted in the ex vivo culture supernatant were determined. We observed a significant increase in the proportion of NK cells producing IFNγ in the group treated with the anti–PD-1/anti-CD96 combination (Fig. 5C). Similarly, the levels of IFNγ secreted into supernatant were also increased in the group treated with the anti–PD-1/anti-CD96 combination (Fig. 5D). Interestingly, an increase in IL2 secretion into the culture supernatant in anti-CD96 and anti-CD96/anti–PD-1-treated groups was detected (Fig. 5E). No significant toxicities were noted in mice treated intensively with anti-CD96, anti–PD-1, or the combination of both (Supplementary Fig. S8A–S8I). Altogether, these data suggest that treatment with anti-CD96 mAb increases immune infiltrates into the lungs of tumor-bearing mice, and the combination of anti-CD96/anti–PD-1 can enhance NK-cell activity.

We previously demonstrated that the incidence of 3-methylcholanthrene (MCA)–induced fibrosarcomas was significantly reduced in Cd96^−/− mice, or in WT mice prophylactically treated with anti-CD96 mAb (32). Therefore, we examined the therapeutic potential of anti-CD96 mAb in treating primary MCA-induced fibrosarcomas. Fibrosarcomas were induced in WT mice by s.c. injection of 300 μg MCA and, once established, treated with mAbs over 6 weeks as described in Materials and Methods. Fibrosarcomas treated with cIg grew rapidly following their development (Fig. 6A), whereas treatment with anti-CD96 or anti–PD-1 significantly reduced the growth of some tumors (Fig. 6B and C). Combining anti-CD96 and anti–PD-1 only marginally reduced the tumor growth rate further compared with each agent as a monotherapy (Fig. 6D and E). However, mice treated with anti-CD96 and anti–PD-1 in combination had a higher rate of tumor rejection, with 3 of 20 mice rejecting tumors compared with 0 of 20 and 1 of 20 in mice treated with anti-CD96 or anti–PD-1, respectively. Although these effects appear modest, they compare well with previous single or combined immunotherapies applied in this model (39, 41). Overall, these data demonstrate that anti-CD96 is also a promising new immunotherapy for treating primary cancers arising de novo.

Metastases stemming from a primary malignancy are a major hurdle in the treatment of surgically resectable cancers, and approaches to limit or treat metastases are of great clinical importance. NK cells play a critical role in the clearance of early metastasis, and strategies to enhance NK-cell activity are emerging as a viable immunotherapeutic approach (42). Interactions between tumor and NK cells leading to cytotoxicity and cytokine production are dependent on the balance of inhibitory and activating receptor interactions. Thus, targeting NK cell–inhibitory receptors to enhance NK-cell function is an attractive immunotherapy. In this study, we have defined a role for the immunoglobulin family receptor CD96 in inhibiting NK cell–mediated control of tumor metastases. Mice lacking CD96 were resistant to experimental lung metastases, and, similarly, treatment with anti-CD96 mAb also enhanced protection against metastases. Control of metastases by anti-CD96 mAb was mediated via NK cells, CD226, and enhanced IFNγ production. We also found that anti-CD96 mAb further inhibited metastases when used in combination with contemporary immune checkpoint inhibitors such as anti–CTLA-4 or anti–PD-1. Indeed, the combination with anti–PD-1 increased NK-cell IFNγ production and infiltration into the lungs of mice bearing B16F10 metastases. Although the effectiveness of treating established B16F10 lung metastases with anti-CD96/anti–PD-1 was reduced compared with mice treated before tumors established, a significant effect of the combination was still observed. Coupled with the effectiveness of anti-CD96 alone or in combination with anti–PD-1 at reducing E0771 or 4T1.2 spontaneous metastases (where adjuvant treatment commences at the point of established metastases), these data suggest anti-CD96 therapy has some activity against nascent metastatic foci.

Following on from the dramatic clinical successes of anti–CTLA-4 (7) and anti–PD-1/PD-L1 (8, 10) mAbs, there has been a rapid exploration of mAbs blocking other immune checkpoints, as well as coactivating receptors in many phase I to III clinical trials (43). In addition, the benefit observed with combining anti–CTLA-4 and anti–PD-1 mAbs (9) suggests that many new immunotherapies will also be tested in combinations. We recently reported Cd96^−/− mice to be highly resistant to experimental lung metastases (32), but had not examined targeting CD96 in a therapeutic context. In this study, we have illustrated the potential clinical utility of targeting CD96 by demonstrating that anti-CD96 mAbs have potent antimetastatic activity. Anti-CD96 mAb inhibited metastasis formation in three different experimental lung metastasis models (of melanoma, prostate cancer, and lung cancer) and was effective as an adjuvant therapy in models of primary tumor resection and spontaneous metastasis. This is an important observation because so far, preclinical data supporting an antimetastatic effect of immune checkpoint blockade are quite sparse. Anti-CD96 mAb also enhanced the antimetastatic activity of the checkpoint inhibitors anti–PD-1 and anti–CTLA-4, as well as the chemotherapeutic effect of DOX. The combination of anti-CD96 and anti–PD-1 had the strongest antimetastatic effect, which is relevant because anti–PD-1 therapy is likely to become a frontline immunotherapy agent based on its safety and efficacy profile across multiple cancer types (10, 11). Of importance, although we were able to demonstrate that anti-CD96/anti–PD-1 increased NK-cell activity in mice, we did not observe any overt signs of illness in these mice, suggesting this combination may not have significant side effects. Ultimately, any immune-related adverse events are best tested in early-phase trials in cancer patients. Although the role of PD-1 in controlling T-cell responses to tumors is well established (44), it can also be expressed on NK cells and alter their activity to tumors (45). Here in the metastatic setting, anti-CD96/anti–PD-1 activity appeared NK cell–dependent. But preliminary experiments in de novo fibrosarcomas suggested the anti-CD96/anti–PD-1 combination might have a broader effect against primary tumors. A combination therapy that might enhance both NK-cell and T-cell functions will be of great clinical utility.

The interactions of NK cell–expressed receptors CD96, CD226, and TIGIT with their ligands, leading to modulated NK-cell activity, are highly complex. However, their emerging role in controlling both NK-cell and T-cell functions suggests their manipulation to be an attractive clinical approach (16). In mice, the blocking of CD226 strongly reduced the activity of anti-CD96 mAb therapy, suggesting a dominant role for CD226 in metastasis control. Interestingly, although Tigit^−/− mice showed no altered resistance to metastases compared with WT mice, the addition of anti-CD96 mAb was able to significantly enhance control of metastases in these mice compared with WT mice. Although TIGIT has been previously identified as a suppressor of NK cells, its expression differs between mice and humans, as resting human NK cells express TIGIT (27), whereas expression is not detectable on mouse NK cells without stimulation (32). As resting mouse NK cells express low levels of TIGIT and high levels of CD96, it is likely that CD96, rather than TIGIT, is more important for NK-cell function at early stages of tumor development, such as experimental metastasis formation within lungs. Nevertheless, the activity of anti-CD96 in Tigit^−/− mice strongly warrants further investigation into co-blockade of these receptors.

Many questions remain to be addressed, including the expression and function of CD96 on human NK cells, other innate lymphoid cells, and T-cell subsets in normal and tumor immunopathology settings. CD96 was reported to be expressed on primary human NK cells (33), and we have confirmed expression on NK-cell and CD4⁺ and CD8⁺ T-cell subsets derived from primary human peripheral blood mononuclear cells (data not shown). Recently, CD96 was identified, with TIGIT, as amongst the top 20 T cell–associated genes expressed in a screen of human lung squamous cell carcinoma samples (38). Human CD96 has been previously reported to play a role in facilitating NK-cell adhesion/cytotoxicity against CD155-expressing tumor cells (33, 46). Whether anti-human CD96 mAbs can promote NK-cell/T-cell function remains to be established. The relative expression and function of CD226, TIGIT, CD96, and their ligands will now be important to characterize in human diseases, notably cancer. Nonetheless, this study has shown CD96 to be an attractive target to enhance NK-cell control of metastasis in mice. The efficacy of anti-CD96 mAb translated across all experimental and spontaneous models of metastases investigated and was more potent than either CTLA-4 or PD-1 blockade. The potent antimetastatic activity of anti-CD96 mAb and its ability to act in combination with anti–CTLA-4, anti–PD-1, or chemotherapy suggest blocking CD96 will be an attractive approach to enhance NK-cell control of metastases.

C57BL/6 and BALB/c WT mice were purchased from the Walter and Eliza Hall Institute for Medical Research or Animal Resource Centre. C57BL/6 Cd96^−/−, Rag1^−/−, Rag2^−/−, and Rag2^−/−;Il2rg^−/− mice were maintained as previously described (32, 34). NKp46^Cre/WT;Mcl1^WT/WT, NKp46^Cre/WT;Mcl1^fl/fl mice were maintained as previously described (47). C57BL/6 Tigit^−/− mice as described previously (32) were kindly provided by Bristol-Myers Squibb. Cd226^−/− and Il12p35^−/− mice have already been described (26, 48). C57BL/6 FcgRIII^−/− and FcγRIV^−/− mice as previously described (49) were kindly provided by Dr. Jeffrey Ravetch. C57BL/6 Fceg^−/− mice as previously described (50) were kindly provided by Professor Mark Hogarth. All mice were bred and maintained at the QIMR Berghofer Medical Research Institute and used between the ages of 6 to 14 weeks. Groups of 5 to 10 mice per experiment were used for experimental tumor metastases. No mice were excluded based on preestablished criteria in this study, and no active randomization was applied to experimental groups. The investigators were not blinded to the group allocation during the experiment and/or when assessing the outcome. All experiments were approved by the QIMR Berghofer Medical Research Institute Animal Ethics Committee.

B16F10 melanoma (ATCC), LWT1 melanoma, 3LL lung carcinoma, 4T1.2 mammary carcinoma, E0771 mammary carcinoma, and RM-1 prostate carcinoma cell lines were maintained, injected, and monitored as previously described (26, 51–53). All cell lines were routinely tested negative for Mycoplasma, but cell line authentication was not routinely performed.

All tumor experiments were performed once unless specifically indicated. Single-cell suspensions of B16F10 melanoma cells (2 × 10⁵), 3LL lung carcinoma cells (1 × 10⁵), RM-1 prostate carcinoma cells (1 × 10⁴), or LWT1 melanoma cells (5 × 10⁵) were injected i.v. into the tail vein of the indicated strains of mice. Lungs were harvested on day 14, and tumor nodules were counted under a dissection microscope.

For spontaneous metastasis and postsurgery survival experiments, 5 × 10⁴ 4T1.2 tumor cells or 2 × 10⁴ E0771 tumor cells were inoculated into the fourth mammary fat pad of BALB/c or C57BL/6 mice, respectively. On the indicated day after injection, mice were anesthetized, the primary tumor surgically removed, and the wound closed with surgical clips before treatment commenced as indicated (DOX, anti–CTLA-4, anti–PD-1, or anti-CD96 alone or in combination). Survival of the mice was monitored, or alternatively at the day indicated after tumor inoculation, mice were sacrificed, lungs were harvested and fixed, and metastatic colonies were counted for individual mice under a dissecting microscope.

Groups of 20 male C57BL/6 WT mice were inoculated s.c. in the hind flank with 300 μg of MCA (Sigma-Aldrich) in 0.1 mL of corn oil as described (51). Mice were treated with cIg, anti-CD96, anti–PD-1, or anti–PD-1/anti-CD96 (250 μg i.p., twice/week) for 6 weeks from the second palpable tumor measurement (∼0.2–0.4 cm², days 84–147 relative to MCA inoculation). Mice were then monitored for fibrosarcoma development over 200 days, with measurements made with a caliper square as the product of two perpendicular diameters (cm²). Data were recorded as tumor size in cm² of individual mice, or tumor growth rate (cm²/day) relative to treatment initiation.

Mice were treated with cIg (2A3 or 1–117), anti-CD96 (3.3, rat IgG1; ref. 32), anti-CD96 (6A6, rat IgG2a; ref. 31), anti–PD-1 (CD279) (RMP1-14, rat IgG2a), anti–CTLA-4 (CD152) (UC10-4F10, hamster IgG, kindly provided by Jeffrey Bluestone), or their combination using schedules and doses as indicated. Anti-CD96 clone 6A6 was used only in Fig. 1E, anti-CD96 mAb clone 3.3 was used for all other experiments, and their affinity for recombinant CD96 was determined as described in Supplementary Methods. Some mice received DOX 2 mg/kg i.v. as indicated and previously described (40). Some mice additionally received either anti-CD4 (GK1.5) and anti-CD8β (53.5.8) as indicated to deplete T-cell subsets; anti-asialoGM1 to deplete NK cells; and anti-IFNγ (H22) or anti-CD226 (480.1) as previously described (32, 39).

Single-cell suspensions were generated from PBS-perfused mouse lungs by mincing lungs with scissors and incubating tissue with Collagenase type IV (Worthington Chemicals) and DNAse I (Roche) in RPMI for 45 minutes at 37ºC. Samples were then passed through a 40 μm cell strainer, washed with PBS, and incubated with 2.4G2 (anti-CD16/32, to block Fc receptors) on ice. Cells were then surface-stained with the following antibodies: anti-CD45.2 (104), anti-TCRβ (H57-597), anti-CD8α (53-6.7), anti-CD4 (RM4-5), anti-NK1.1 (PK136), anti-NKp46 (29A1.4), anti-CD96 (3.3), anti-CD155 (TX56), anti-CD122 (829038), and live/dead dye Zombie Aqua (all from BioLegend, eBioscience, or R&D Systems). To stain for intracellular IFNγ, cells were surface-stained as described above before being fixed/permeabilized with a cytofix/cytoperm kit (BD Biosciences) and stained with anti-IFNγ (XMG1.2) or respective isotype (BioLegend). To determine absolute counts in samples, liquid-counting beads (BD Biosciences) were added directly before samples were run on a flow cytometer. All data were collected on a Fortessa 4 (BD) flow cytometer and analyzed with FlowJo v10 software (Tree Star, Inc.).

Single-cell suspensions were generated from PBS-perfused mouse lungs. Cells were counted and resuspended in complete RPMI supplemented with 10% FCS (Thermo Scientific), penicillin–streptomycin, l-glutamine, nonessential amino acids, sodium pyruvate, and HEPES (GIBCO) to 5 × 10⁶ cells/ml and 200 μL/well added to a 96-U-bottom plate. Cells were incubated at 37°C for 24 hours before supernatants were collected, and cells stained for surface markers and intracellular IFNγ as described above. Cytokine levels in supernatants were determined using cytokine bead array as per the manufacturer's instructions (BD Biosciences).

Statistical analysis was achieved using Graphpad Prism Software. Data were considered to be statistically significant where the P value was equal to or less than 0.05. Data were compared using a Mann–Whitney U test. Differences in survival were evaluated using a log-rank test.

J.J. Miles reports receiving a commercial research grant from Bristol-Myers Squibb. G. Bernhardt is a consultant/advisory board member for MorphoSys. M.J. Smyth reports receiving a commercial research grant from Bristol-Myers Squibb and other commercial research support from MedImmune. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.J. Blake, M.J. Smyth

Development of methodology: K. Stannard, A. Roman Aguilera, L. Ferrari de Andrade, M.J. Smyth

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.J. Blake, K. Stannard, J. Liu, M.C.R. Yong, D. Mittal, J.J. Miles, V.P. Lutzky, L. Ferrari de Andrade, M. Colonna, G. Bernhardt, M.J. Smyth

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.J. Blake, A. Roman Aguilera, J.J. Miles, V.P. Lutzky, L. Ferrari de Andrade, M.W.L. Teng, M.J. Smyth

Writing, review, and/or revision of the manuscript: S.J. Blake, S. Allen, A. Roman Aguilera, J.J. Miles, V.P. Lutzky, L. Ferrari de Andrade, L. Martinet, F. Kühnel, E. Gurlevik, M.W.L. Teng, M.J. Smyth

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Stannard, J. Liu, S. Allen, M.C.R. Yong, A. Roman Aguilera, K. Takeda

Study supervision: M.W.L. Teng, M.J. Smyth

The authors thank Liam Town, Kate Elder, and Joanne Sutton for breeding, genotyping, and maintenance and care of the mice used in this study. They also thank Jeffrey Ravetch for providing the original C57BL/6 FcγR III and FcγR IV gene-targeted breeding pairs, Mark Hogarth for providing the original C57BL/6 Fceγ gene-targeted breeding pairs, and Bristol-Myers Squibb for providing the original C57BL/6 TIGIT gene-targeted breeding pairs.

The project was funded by a National Health and Medical Research Council of Australia (NH&MRC) Project Grant (1044392) and Development Grant (1093566), a Cancer Council of Queensland (CCQ) Project Grant (1083776), and a Cancer Research Institute CLIP Grant. M.J. Smyth is supported by a Senior Principal Research Fellowship (1078671). M.W.L. Teng is supported by a CDF1 Fellowship and Project Grant from NH&MRC, a Prostate Cancer Foundation of Australia Grant, and a CCQ Project Grant. L. Ferrari de Andrade was supported by a Conselho Nacional de Desenvolvimento Científico e Tecnológico PhD Fellowship.